Datasets:
NimaBoscarino
commited on
Commit
•
ac9a2bd
1
Parent(s):
ba39dc9
Update LILA.py
Browse files
LILA.py
CHANGED
@@ -53,31 +53,30 @@ _LICENSE = ""
|
|
53 |
_LILA_SAS_URLS = pd.read_csv("https://lila.science/wp-content/uploads/2020/03/lila_sas_urls.txt")
|
54 |
_LILA_SAS_URLS.rename(columns={"# name": "name"}, inplace=True)
|
55 |
|
|
|
|
|
56 |
# How do I make these point to the particular commit ID?
|
57 |
_LILA_URLS = {
|
58 |
-
"Caltech Camera Traps": "
|
59 |
-
"ENA24": "
|
60 |
-
"Missouri Camera Traps": "",
|
61 |
-
"NACTI": "",
|
62 |
-
"WCS Camera Traps": "",
|
63 |
-
"Wellington Camera Traps": "",
|
64 |
-
"Island Conservation Camera Traps": "",
|
65 |
-
"Channel Islands Camera Traps": "",
|
66 |
-
"Idaho Camera Traps": "",
|
67 |
-
"Snapshot Serengeti": "",
|
68 |
-
"Snapshot Karoo": "",
|
69 |
-
"Snapshot Kgalagadi": "",
|
70 |
-
"Snapshot Enonkishu": "",
|
71 |
-
"Snapshot Camdeboo": "",
|
72 |
-
"Snapshot Mountain Zebra": "",
|
73 |
-
"Snapshot Kruger": "",
|
74 |
-
"SWG Camera Traps": "",
|
75 |
-
"Orinoquia Camera Traps": "",
|
76 |
}
|
77 |
|
78 |
-
# TODO: Just to make the Dataset viewer on the Hub work
|
79 |
-
DEFAULT_CONFIG_NAME = "Caltech Camera Traps"
|
80 |
-
|
81 |
class LILAConfig(datasets.BuilderConfig):
|
82 |
"""Builder Config for LILA"""
|
83 |
|
@@ -101,7 +100,7 @@ class LILA(datasets.GeneratorBasedBuilder):
|
|
101 |
name=row.name,
|
102 |
# description="TODO: Description",
|
103 |
image_base_url=row.image_base_url,
|
104 |
-
metadata_url=_LILA_URLS[row.name]
|
105 |
) for row in _LILA_SAS_URLS.itertuples()
|
106 |
]
|
107 |
|
@@ -119,19 +118,15 @@ class LILA(datasets.GeneratorBasedBuilder):
|
|
119 |
"location": datasets.Value("string"),
|
120 |
"rights_holder": datasets.Value("string"),
|
121 |
"frame_num": datasets.Value("int32"),
|
122 |
-
|
123 |
-
|
124 |
"annotations": datasets.Sequence({
|
125 |
"id": datasets.Value("string"),
|
126 |
"category_id": datasets.Value("int32"),
|
127 |
}),
|
128 |
-
|
129 |
"bboxes": datasets.Sequence({
|
130 |
"id": datasets.Value("string"),
|
131 |
"category_id": datasets.Value("int32"),
|
132 |
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
133 |
}),
|
134 |
-
|
135 |
"image": datasets.Image(decode=False),
|
136 |
})
|
137 |
elif self.config.name == 'ENA24':
|
@@ -145,6 +140,192 @@ class LILA(datasets.GeneratorBasedBuilder):
|
|
145 |
}),
|
146 |
"image": datasets.Image(decode=False),
|
147 |
})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
148 |
|
149 |
def _info(self):
|
150 |
features = self._get_features()
|
@@ -167,6 +348,8 @@ class LILA(datasets.GeneratorBasedBuilder):
|
|
167 |
|
168 |
def _split_generators(self, dl_manager):
|
169 |
archive_path = dl_manager.download_and_extract(self.config.metadata_url)
|
|
|
|
|
170 |
|
171 |
return [
|
172 |
datasets.SplitGenerator(
|
|
|
53 |
_LILA_SAS_URLS = pd.read_csv("https://lila.science/wp-content/uploads/2020/03/lila_sas_urls.txt")
|
54 |
_LILA_SAS_URLS.rename(columns={"# name": "name"}, inplace=True)
|
55 |
|
56 |
+
_METADATA_BASE_URL = "https://huggingface.co/datasets/NimaBoscarino/LILA/resolve/main/data/"
|
57 |
+
|
58 |
# How do I make these point to the particular commit ID?
|
59 |
_LILA_URLS = {
|
60 |
+
"Caltech Camera Traps": "Caltech_Camera_Traps.jsonl",
|
61 |
+
"ENA24": "ENA24.jsonl",
|
62 |
+
"Missouri Camera Traps": "Missouri_Camera_Traps.jsonl",
|
63 |
+
"NACTI": "NACTI.jsonl.zip",
|
64 |
+
"WCS Camera Traps": "WCS_Camera_Traps.jsonl.zip",
|
65 |
+
"Wellington Camera Traps": "Wellington_Camera_Traps.jsonl.zip",
|
66 |
+
"Island Conservation Camera Traps": "Island_Conservation_Camera_Traps.jsonl.zip",
|
67 |
+
"Channel Islands Camera Traps": "Channel_Islands_Camera_Traps.jsonl.zip",
|
68 |
+
"Idaho Camera Traps": "Idaho_Camera_Traps.jsonl.zip",
|
69 |
+
"Snapshot Serengeti": "Snapshot_Serengeti.jsonl.zip",
|
70 |
+
"Snapshot Karoo": "Snapshot_Karoo.jsonl.zip",
|
71 |
+
"Snapshot Kgalagadi": "Snapshot_Kgalagadi.jsonl",
|
72 |
+
"Snapshot Enonkishu": "Snapshot_Enonkishu.jsonl.zip",
|
73 |
+
"Snapshot Camdeboo": "Snapshot_Camdeboo.jsonl.zip",
|
74 |
+
"Snapshot Mountain Zebra": "Snapshot_Mountain_Zebra.jsonl.zip",
|
75 |
+
"Snapshot Kruger": "Snapshot_Kruger.jsonl",
|
76 |
+
"SWG Camera Traps": "SWG_Camera_Traps.jsonl.zip",
|
77 |
+
"Orinoquia Camera Traps": "Orinoquia_Camera_Traps.jsonl.zip",
|
78 |
}
|
79 |
|
|
|
|
|
|
|
80 |
class LILAConfig(datasets.BuilderConfig):
|
81 |
"""Builder Config for LILA"""
|
82 |
|
|
|
100 |
name=row.name,
|
101 |
# description="TODO: Description",
|
102 |
image_base_url=row.image_base_url,
|
103 |
+
metadata_url=_METADATA_BASE_URL + _LILA_URLS[row.name]
|
104 |
) for row in _LILA_SAS_URLS.itertuples()
|
105 |
]
|
106 |
|
|
|
118 |
"location": datasets.Value("string"),
|
119 |
"rights_holder": datasets.Value("string"),
|
120 |
"frame_num": datasets.Value("int32"),
|
|
|
|
|
121 |
"annotations": datasets.Sequence({
|
122 |
"id": datasets.Value("string"),
|
123 |
"category_id": datasets.Value("int32"),
|
124 |
}),
|
|
|
125 |
"bboxes": datasets.Sequence({
|
126 |
"id": datasets.Value("string"),
|
127 |
"category_id": datasets.Value("int32"),
|
128 |
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
129 |
}),
|
|
|
130 |
"image": datasets.Image(decode=False),
|
131 |
})
|
132 |
elif self.config.name == 'ENA24':
|
|
|
140 |
}),
|
141 |
"image": datasets.Image(decode=False),
|
142 |
})
|
143 |
+
elif self.config.name == 'Missouri Camera Traps':
|
144 |
+
return datasets.Features({
|
145 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
146 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
147 |
+
"seq_id": datasets.Value("string"), "seq_num_frames": datasets.Value("int32"),
|
148 |
+
"frame_num": datasets.Value("int32"),
|
149 |
+
"annotations": datasets.Sequence({
|
150 |
+
"id": datasets.Value("string"),
|
151 |
+
"category_id": datasets.Value("int32"),
|
152 |
+
"sequence_level_annotation": datasets.Value("bool"),
|
153 |
+
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
154 |
+
}),
|
155 |
+
"image": datasets.Image(decode=False),
|
156 |
+
})
|
157 |
+
elif self.config.name == 'NACTI':
|
158 |
+
return datasets.Features({
|
159 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
160 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
161 |
+
"study": datasets.Value("string"), "location": datasets.Value("string"),
|
162 |
+
"annotations": datasets.Sequence({
|
163 |
+
"id": datasets.Value("string"),
|
164 |
+
"category_id": datasets.Value("int32"),
|
165 |
+
}),
|
166 |
+
"bboxes": datasets.Sequence({
|
167 |
+
"id": datasets.Value("string"),
|
168 |
+
"category_id": datasets.Value("int32"),
|
169 |
+
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
170 |
+
}),
|
171 |
+
"image": datasets.Image(decode=False),
|
172 |
+
})
|
173 |
+
elif self.config.name == 'WCS Camera Traps':
|
174 |
+
return datasets.Features({
|
175 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
176 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
177 |
+
"wcs_id": datasets.Value("string"), "location": datasets.Value("string"),
|
178 |
+
"frame_num": datasets.Value("int32"), "match_level": datasets.Value("int32"),
|
179 |
+
"seq_id": datasets.Value("string"), "country_code": datasets.Value("string"),
|
180 |
+
"seq_num_frames": datasets.Value("int32"),
|
181 |
+
"status": datasets.Value("string"),
|
182 |
+
"datetime": datasets.Value("date32"),
|
183 |
+
"corrupt": datasets.Value("bool"),
|
184 |
+
"annotations": datasets.Sequence({
|
185 |
+
"id": datasets.Value("string"),
|
186 |
+
"category_id": datasets.Value("int32"),
|
187 |
+
"count": datasets.Value("int32"),
|
188 |
+
"sex": datasets.Value("string"),
|
189 |
+
"age": datasets.Value("string"),
|
190 |
+
}),
|
191 |
+
"bboxes": datasets.Sequence({
|
192 |
+
"id": datasets.Value("string"),
|
193 |
+
"category_id": datasets.Value("int32"),
|
194 |
+
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
195 |
+
}),
|
196 |
+
"image": datasets.Image(decode=False),
|
197 |
+
})
|
198 |
+
elif self.config.name == 'Wellington Camera Traps':
|
199 |
+
return datasets.Features({
|
200 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
201 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
202 |
+
"frame_num": datasets.Value("int32"), "seq_id": datasets.Value("string"),
|
203 |
+
"site": datasets.Value("string"), "camera": datasets.Value("string"),
|
204 |
+
"datetime": datasets.Value("date32"),
|
205 |
+
"annotations": datasets.Sequence({
|
206 |
+
"id": datasets.Value("string"),
|
207 |
+
"category_id": datasets.Value("int32"),
|
208 |
+
}),
|
209 |
+
"image": datasets.Image(decode=False),
|
210 |
+
})
|
211 |
+
elif self.config.name == 'Island Conservation Camera Traps':
|
212 |
+
return datasets.Features({
|
213 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
214 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
215 |
+
"annotations": datasets.Sequence({
|
216 |
+
"id": datasets.Value("string"),
|
217 |
+
"category_id": datasets.Value("int32"),
|
218 |
+
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
219 |
+
}),
|
220 |
+
"image": datasets.Image(decode=False),
|
221 |
+
})
|
222 |
+
elif self.config.name == 'Channel Islands Camera Traps':
|
223 |
+
return datasets.Features({
|
224 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
225 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
226 |
+
"frame_num": datasets.Value("int32"), "seq_id": datasets.Value("string"),
|
227 |
+
"seq_num_frames": datasets.Value("int32"),
|
228 |
+
"original_relative_path": datasets.Value("string"),
|
229 |
+
"location": datasets.Value("string"),
|
230 |
+
"temperature": datasets.Value("string"),
|
231 |
+
"annotations": datasets.Sequence({
|
232 |
+
"id": datasets.Value("string"),
|
233 |
+
"category_id": datasets.Value("int32"),
|
234 |
+
"sequence_level_annotation": datasets.Value("bool"),
|
235 |
+
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
236 |
+
}),
|
237 |
+
"image": datasets.Image(decode=False),
|
238 |
+
})
|
239 |
+
elif self.config.name == 'Idaho Camera Traps':
|
240 |
+
return datasets.Features({
|
241 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
242 |
+
"frame_num": datasets.Value("int32"), "seq_id": datasets.Value("string"),
|
243 |
+
"seq_num_frames": datasets.Value("int32"),
|
244 |
+
"original_relative_path": datasets.Value("string"),
|
245 |
+
"datetime": datasets.Value("date32"),
|
246 |
+
"location": datasets.Value("string"),
|
247 |
+
"annotations": datasets.Sequence({
|
248 |
+
"id": datasets.Value("string"),
|
249 |
+
"category_id": datasets.Value("int32"),
|
250 |
+
"sequence_level_annotation": datasets.Value("bool"),
|
251 |
+
}),
|
252 |
+
"image": datasets.Image(decode=False),
|
253 |
+
})
|
254 |
+
elif self.config.name == 'Snapshot Serengeti':
|
255 |
+
return datasets.Features({
|
256 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
257 |
+
"frame_num": datasets.Value("int32"), "seq_id": datasets.Value("string"),
|
258 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
259 |
+
"seq_num_frames": datasets.Value("int32"),
|
260 |
+
"datetime": datasets.Value("date32"),
|
261 |
+
"corrupt": datasets.Value("bool"),
|
262 |
+
"location": datasets.Value("string"),
|
263 |
+
"annotations": datasets.Sequence({
|
264 |
+
"id": datasets.Value("string"),
|
265 |
+
"category_id": datasets.Value("int32"),
|
266 |
+
"sequence_level_annotation": datasets.Value("bool"),
|
267 |
+
"seq_id": datasets.Value("string"),
|
268 |
+
"season": datasets.Value("string"),
|
269 |
+
"datetime": datasets.Value("date32"),
|
270 |
+
"subject_id": datasets.Value("string"),
|
271 |
+
"count": datasets.Value("string"),
|
272 |
+
"standing": datasets.Value("float32"),
|
273 |
+
"resting": datasets.Value("float32"),
|
274 |
+
"moving": datasets.Value("float32"),
|
275 |
+
"interacting": datasets.Value("float32"),
|
276 |
+
"young_present": datasets.Value("float32"),
|
277 |
+
"location": datasets.Value("string"),
|
278 |
+
}),
|
279 |
+
"bboxes": datasets.Sequence({
|
280 |
+
"id": datasets.Value("string"),
|
281 |
+
"category_id": datasets.Value("int32"),
|
282 |
+
"bbox": datasets.Sequence(datasets.Value("float32"), length=4),
|
283 |
+
}),
|
284 |
+
"image": datasets.Image(decode=False),
|
285 |
+
})
|
286 |
+
elif self.config.name in [
|
287 |
+
'Snapshot Karoo', 'Snapshot Kgalagadi', 'Snapshot Enonkishu', 'Snapshot Camdeboo',
|
288 |
+
'Snapshot Mountain Zebra', 'Snapshot Kruger'
|
289 |
+
]:
|
290 |
+
return datasets.Features({
|
291 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
292 |
+
"frame_num": datasets.Value("int32"), "seq_id": datasets.Value("string"),
|
293 |
+
"width": datasets.Value("int32"), "height": datasets.Value("int32"),
|
294 |
+
"seq_num_frames": datasets.Value("int32"),
|
295 |
+
"datetime": datasets.Value("date32"),
|
296 |
+
"corrupt": datasets.Value("bool"),
|
297 |
+
"location": datasets.Value("string"),
|
298 |
+
"annotations": datasets.Sequence({
|
299 |
+
"id": datasets.Value("string"),
|
300 |
+
"category_id": datasets.Value("int32"),
|
301 |
+
"sequence_level_annotation": datasets.Value("bool"),
|
302 |
+
"seq_id": datasets.Value("string"),
|
303 |
+
"season": datasets.Value("string"),
|
304 |
+
"datetime": datasets.Value("date32"),
|
305 |
+
"subject_id": datasets.Value("string"),
|
306 |
+
"count": datasets.Value("string"),
|
307 |
+
"standing": datasets.Value("float32"),
|
308 |
+
"resting": datasets.Value("float32"),
|
309 |
+
"moving": datasets.Value("float32"),
|
310 |
+
"interacting": datasets.Value("float32"),
|
311 |
+
"young_present": datasets.Value("float32"),
|
312 |
+
"location": datasets.Value("string"),
|
313 |
+
}),
|
314 |
+
"image": datasets.Image(decode=False),
|
315 |
+
})
|
316 |
+
elif self.config.name == 'Orinoquia Camera Traps':
|
317 |
+
return datasets.Features({
|
318 |
+
"id": datasets.Value("string"), "file_name": datasets.Value("string"),
|
319 |
+
"frame_num": datasets.Value("int32"), "seq_id": datasets.Value("string"),
|
320 |
+
"seq_num_frames": datasets.Value("int32"), "datetime": datasets.Value("date32"),
|
321 |
+
"location": datasets.Value("string"),
|
322 |
+
"annotations": datasets.Sequence({
|
323 |
+
"id": datasets.Value("string"),
|
324 |
+
"sequence_level_annotation": datasets.Value("bool"),
|
325 |
+
"category_id": datasets.Value("int32"),
|
326 |
+
}),
|
327 |
+
"image": datasets.Image(decode=False),
|
328 |
+
})
|
329 |
|
330 |
def _info(self):
|
331 |
features = self._get_features()
|
|
|
348 |
|
349 |
def _split_generators(self, dl_manager):
|
350 |
archive_path = dl_manager.download_and_extract(self.config.metadata_url)
|
351 |
+
if archive_path.endswith(".zip"):
|
352 |
+
archive_path = os.path.join(archive_path, os.listdir(archive_path)[0])
|
353 |
|
354 |
return [
|
355 |
datasets.SplitGenerator(
|